Till startsida
Till innehåll Läs mer om hur kakor används på gu.se


Mediation analysis to quantify direct and indirect effect of smoking

Much of the research in epidemiology is concerned with understanding the causal pathways between exposure and outcome.

In a project started 2017, we use mediation analysis to quantify the direct effect of smoking on cardio-vascular events vs the indirect effect, mediated through higher blood cadmium levels. The mediation analysis is performed using regression analysis to estimate both the smoking/cvd association and the smoking/cadmium association (sometimes called the product method). Different regression methods are used depending on how the health outcome is measured (prevalence, time to event), how the exposure is measured (smoking status, packyears) and the distribution of the mediator (normal, skewed). Within this project, computer procedures for these different situations will be developed. The inference is mainly based on bootstrap resampling.

Using data from a Swedish cohort, Malmö Diet and Cancer, we could show that 60-65% of the association between atherosclerotic plaques and smoking is mediated through high blood cadmium levels. In these projects we collaborate with researchers from Malmö (prof Gunnar Engström).

Regression analysis for a skewed distribution

A skewed distribution for the exposure is often taken to be approximately log-normal. In statistical analysis and inference, a normal distribution if often assumed, and therefore the exposure is log-transformed. However, the log-transformation implies that the determinants have a multiplicative association with the exposure (e.g. the expected geometric mean for benzene among smokers is 49% higher than that of non-smokers, after adjusting for other determinants).

A new regression method for estimating the additive association with the arithmetic mean has been developed.

View our publications on

--> Biostatistics

Sidansvarig: Madeleine Modig|Sidan uppdaterades: 2017-10-30

På Göteborgs universitet använder vi kakor (cookies) för att webbplatsen ska fungera på ett bra sätt för dig. Genom att surfa vidare godkänner du att vi använder kakor.  Vad är kakor?